Chapter 4 SUPERHELICAL DNA STUDIED BY SOLUTION SCATTERING AND COMPUTER MODELS

Jrg Langowski, Markus Hammermann, Konstantin Klenin and Katalin Tth

Deutsches Krebsforschungszentrum, Division Biophysics of Macromolecules (H0500), Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany E-mail: joerg.langowski@dkfz-heidelberg.de

Abstract:

We present here recent results on the structure of superhelical DNA and its changes with salt concentration between 0.01 and 1.5 M NaCl. Scattering curves of two different superhelical DNAs were determined by static light scattering. The measured radii of gyration do not change significantly with salt concentration. Small-angle neutron scattering, together with calculations from a Monte-Carlo model, allows to determine the superhelix diameter. Measured and simulated scattering curves agreed almost quantitatively. Experimentally we find that the diameter decreases from (16.00.9) nm at 10 mM to (9.00.7) nm at 100 mM NaCl. The superhelix diameter from the simulated conformations decreased from (18.01.5) nm at 10 mM to (9.41.5) nm at 100 mM NaCl. At higher salt concentrations up to 1.5 M NaCl, the diameter stays constant at 9 nm.

1.

INTRODUCTION

The compaction of linear DNA under torsional stress to form a superhelix (see Fig. 1) plays a very important role in the packing of DNA in the cell and in regulation of gene activity. In order to understand how superhelicity is related to the biological function of DNA it is important to understand the structure of the superhelix and its changes in different environments. DNA of several kb length is a highly flexible molecule whose structure at room temperature can only be described by an ensemble of molecules of widely varying conformation which all have approximately the same internal energy (Langowski, et al., 1996). The structure of such a 1

2 molecule is best determined free in solution by methods such as solution scattering that disturb the conformational equilibrium as little as possible. Using such techniques one can determine average structural and dynamic properties of the molecule, such as radius of gyration, small angle scattering functions, diffusion coefficients, or parameters related to internal bending and twisting motion, and their dependence on parameters like salt concentration or superhelix density. In order to relate those solution properties to underlying physical properties of the DNA chain, like bending, twisting and stretching elasticity, DNA radius and electrostatic interactions, one has to apply computer modeling techniques.

Fig. 1: Schematic representation of a superhelical DNA (pUC 18, 2686 bp) from a MonteCarlo simulation. The superhelix diameter is defined as the average distance between opposing DNA double strands in the interwound regions.

For modeling a large macromolecule like DNA, atomic level molecular modeling techniques fail because of the large number of atoms involved (10 6 or more) and the simulation time scales required (s to ms). Therefore, a variety of simulation techniques have been developed in which the DNA chain is approximated by a homogeneous elastic filament. We use either a

3 Monte-Carlo (MC) or Brownian dynamics (BD) technique based on a representation of the DNA chain as a segmented flexible chain (Chirico and Langowski, 1996; Klenin et al., 1995; Klenin et al., 1998). One biologically very important and as yet poorly understood problem is the interaction between opposing double strands in an interwound superhelix and its modulation by the concentration of counterions. Recent studies by cryo-electron microscopy (Bednar, et al., 1994) indicated a lateral collapse of the interwound structure for Na+ concentrations > 0.1 M or at millimolar concentrations of Mg 2+. However, Gebe et al. (Gebe, et al., 1996), could not find evidence for such a collapse in static and dynamic light scattering (SDLS), fluorescence polarization anisotropy decay (FPA) or circular dichroism (CD) meas urements. Also, DNA catenation studies by (Rybenkov et al., 1997b) indicate an open structure of the superhelix even at higher ionic strength. Thus, one of the questions that we are dealing with here is whether high salt concentrations can induce a collapse of the superhelical DNA where the two double strands are directly touching each other.

2.

METHODS

Superhelical DNA preparation. pUC18 plasmid DNA (2868 bp) was prepared from E. coli HB 101 as described (Kapp and Langowski, 1992; Langowski, 1987), p1868 (1868 bp) was prepared as described in (Hammermann et al., 1997). After ethanol precipitation and short drying the crude DNA was dissolved in TE-buffer (10mM Tris-HCl, pH 7.5, 1mM EDTA). For additional purif ication and concentration the DNA was precipitated for 15 h on ice by adding a 50% solution of PEG 20000 in 0.5M NaCl to a final PEG concentration of 10% in 0.6M NaCl. The DNA was pelleted at 8000 rpm and 4C in the HB4 rotor in a Sorvall RC-5B centrifuge and again dissolved in TE buffer. Supercoiled and relaxed plasmids were separated by HPLC as described in (Kapp and Langowski, 1992). The integrity of the DNA plasmids and the superhelical density was checked by agarose gel electrophoresis (1%) in Tris-acetate buffer (40 mM Tris.acetate, 2 mM EDTA, pH 8). Only samples containing more than 90% supercoiled DNA were used for neutron scattering measurements in 10mM Tris-HCl, 0.1mM EDTA, pH=7.5 with varying amounts of NaCl. As a control, scattering functions of relaxed plasmids were also collected. The NaCl concentration was varied between 0 and 1.5 M. DNA concentrations were 26 mg/ml. Both plasmids were also dialyzed against the same buffer with 100% D2O, dialysis was performed in 100x sample v olume in Sartorius collodium bags (cellulose nitrate, 12000 D MWCO, FRG) with at least 3 buffer changes.

4 Neutron scattering measurements were performed at the D22 large scale structure diffractometer at the Institut Laue-Langevin (ILL), Grenoble. The sample cuvette (quartz, 1 mm light path, 7 x 10 mm2 illuminated crosssection, Hellma GmbH, Mllheim, FRG) was thermostated at 20 C. The neutron wavelength was 8 , the sample-detector distance was set to 5.0 m or 2.5 m to reach a momentum transfer q range of 0.13 nm-1. Data was collected for at least 1-2 h at each Na+ concentration. After radial integration the data was corrected for sample transmission at 0 and 100% D 2O. Background and buffer scattering was subtracted. For comparison of data sets at different sample concentrations and slight baseline deviations in the D2O samples, the data was renormalized to the same intensity at low q and the baseline at high q was corrected to the same value. Light scattering measurements. Plasmid DNA was filtered slowly (1-2 ml/min) through 0.1m polycarbonate membrane filters (Nuclepore, Pleasanton, Ca, USA), SV40-DNA through 0.2 m respectively 0.4 m membrane filters (Whatman, Maidstone, England) into a cylindrical quartz scattering cell of 1 cm diameter (Hellma, Mlheim, FRG) which had been rinsed with at least 10 ml of filtered buffer before use. Light scattering measurements were performed with vertically polarized argon laser light at 488 nm (Spectra-Physics 2020). Typical power used was 500 mW in TEM00. The sample cuvette was in the center of a cylindrical index matching bath filled with water of 20 C. The scattered light was detected by a photomultiplier (Thorn EMI PM28B for p1868, ALV SOSIPD Dual, Langen, FRG for SV40-DNA) on an ALV goniometer arm. The correlator was a 288 channel ALV-5000 multiple tau digital correlator, Langen, FRG. After careful filtering, dust particles were observed only occasionally (about once every ten seconds) over the length of the laser beam traversing the scattering cell. The length of the observed scattering volume was about 200 m, hence the probability of a dust particle entering this volume was of the order of once every few minutes. Stray dust events were completely eliminated by a 'software dust filter' which collected the data in batches and rejected any data where dust contamination was suspected (see below). The static light scattering measurements yielded the angle-dependent Rayleigh ratio

RR ?? ??

IS (? ) 2 r ? Ko c M P?? ? I0

[1]

where Is(?? is the scattering intensity as measured by the photomultiplier, I0 the intensity of the incident light as measured by a photodiode, r the

5 distance between the scattering volume and the detector, ? the scattering angle, c the concentration and M the molecular weight of th e molecules, P(?? the form factor, and K an optical constant. To calibrate the instrument, a 0 standard measurement was done with toluene, we used a standard Rayleigh ratio of 2.94210-5 cm -1 at 25 C as given by ALV, Langen.

3. 3.1

RESULTS Solution structure of the superhelix

The solution structure of superhelical plasmid DNA was studied by light scattering and small angle neutron scattering. These two methods complement each other since light scattering yields infor mation about the global shape of the molecule (as given by the radius of gyration), while neutron scattering, with its shorter wavelength, probes internal details of the structure (see Fig. 2).

Fig. 2: Typical shape of the scattering form factor of a superhelical DNA in the light scattering (LS) and small angle neutron scattering (SANS) regimes.

The information from solution scattering is rather indirect: one obtains the scattered intensity I as a function of the scattering vector q (defined as q = 4pn/? ?sin(?/2), where ? is the wavelength, ? the scattering angle and n the refractive index of the medium). As mentioned in the introduction, this information can be connected with physical properties of the DNA by a

6 suitable model which can predict the quantities measured in the scattering experiment. Part of the following is a summary of some of the data given in (Hammermann et al., 1998); for more details see the original publication. Light scattering. The structure factor P(q) is defined as the scattered intensity I(q) normalized to ts value extrapolated to zero scattering angle, i P(q) = I(q)/I(0). The structure factor P(q) of superhelical DNA can be well approximated by that of a Gaussian random coil, for which an analytical expression is known:

q2 ? D 2 ? P( q) ? 22 ?exp?? a?? a ? 1? ; a ? [2] a 6 2 where D is the mean squared end-to-end distance of the coil. The D2 radius of gyration RG is related to D 2 : RG = [3]
6

2 Fig. 3: Light scattering intensity as a function of the squared scattering vector q for pUC18 DNA. Symbols: experimental data, line: theoretical form factor for a Gaussian chain (eq. 1).

Fig. 3 shows a typical light scattering curve for pUC18 plasmid DNA at its native superhelical density. From a fit of eq. 2 against the measured data, the apparent radius of gyration could be obtained. The radii of gyration of different plasmids as a function of Na+-concentration are shown in table 1.

7
Table 1: experimental and simulated radii of gyration as a function of NaCl concentration for three different superhelical DNAs: p1868 (1868 bp), pUC18 (2686 bp) and SV40 (5243 bp).

The slight decrease observed is not significant within experimental error, thus any compaction of the superhelix due to screening of the negatively charged DNA backbone by the Na+ counterions is likely to be small. The simulated static structure factor from the last 500 ? s of a Brownian dynamics trajectory can also be fitted very satisfactorily to the experimental data (not shown).

Fig. 4: Small angle neutron scattering curves from pUC18 solutions in H2O buffer, 10mM Tris, at different Na + concentrations: 0 mM (? ), 10 mM (? ), 40 mM ( ?), 100 mM ( ), 500 mM ( ).

Small angle neutron scattering. The measur ed static form factor of the two plasmids p1868 and pUC18 shows the same Na+-concentration dependent behavior: An undulation in the scat tering curve at a momentum transfer q = 0.5 nm -1 seen at 10 mM Tris, 0 mM Na+ shifts to higher q values with

8 increasing Na+ concentration up to 500 mM (Fig. 4). At 1 M and 1.5 M the scattering curve super imposes with the curve at 0.5 M Na+ (data not shown); there was no indication for any further structural change above 100 mM Na+ concentration. For quantifying this effect, we computed the ratio of the scattering intensities of the superhelical DNA and the relaxed form (Fig. 5). The shape of this curve is similar to the form factor of a pair of point scatterers at a constant distance d,

?? sin ? ??? qd f p ? ? ? ?? ? d,q 1 ?? qd ??

2

(solid lines in Fig. 5).

Fig. 5: Ratio of the measured scattering intensities I(q) of pUC18, supercoiled, in H2O, 10mM Tris, at 0 mM (? ) and 100 mM ( ) Na+ concentration and relaxed pUC18 DNA at 100 mM Na+ concentration. The solid lines are the scattering form factors of a pair of point scatterers at a distance r=16.0 nm (thin line) and r=9.0 nm (thick line).

This behavior is to be expected if a certain intramolecular distance occurs with high probability in the superhelical, but not in the relaxed DNA.

9 The Debye formula for the scattering form factor of a macromolecule consisting of identical subunits is

1 P(q) ? 2 N ??

?? ?? ?? sin q r ? r ?? ?? i j ?? ?1 ? 1 ?? ?? f 2 (q) i? j? qr ? r i j
N N

[4]

where f(q) is the q-dependent scattering form factor of the subunit (see e.g. Cantor and Schimmel, 1980). We now compare two macromolecules A and B: A is a chain of point scatterers, B is the same chain but with pairs of point scatterers at a distance d arranged with their center of mass on the backbone ?? of the chain. Thus, in eq. (4) the coordinates r i will be the same for the two ?? structures, only f(q) will be equal to one for structure A and equal to fP(d,q) for structure B. The ratio of the scattering intensities IB/IA will then be fP(d,q). In the case of a superhelical DNA, where a certain intramolecular distance occurs with high probability we should find a very similar behavior. From a fit of the form factor fp(d,q) to the curves in Fig. 5 one can obtain the value of d, which can be regarded as the interstrand sepa ration in interwound regions of the DNA superhelix, i.e., the superhelix diameter as defined in Fig. 1. The depen dence of d on NaCl concentration is shown in Fig. 6; it decreased from 16.00.9 nm at 0 mM to 9.00.8 nm at 100 mM. No further change of d at 500, 1000 or 1500 mM NaCl was detected.

Fig. 6: Measured (? ) and simulated (? ? superhelix diameter of pUC18 vs. salt concentration. ) The values calculated by (Rybenkov et al., 1997) are shown for comparison (?).

10 DNA simulations. The salt-dependent change in the static form factor was predicted in simulated scattering curves of pUC18 and p1868. Super imposed measured and simulated scattering curves of pUC18 are shown in Fig. 7.

Fig. 7: Measured and simulated neutron scattering intensity I(q) of pUC18 in D2O buffer, at 10 mM (Exp.: ? ,? Sim.: solid line) and at 100 mM (Exp.: , Sim.: dashed line) salt concentration.

Since the simulated scattering functions agree very closely with the measured ones for both plasmids, we can assume that the model describes the structure satisfactorily, and fur ther calculations from the simulated conformations are possible which reflect structural features of the DNA in free solution. We calculated the average distribution function pn(r) of each ?? ?? segment r i and its nearest neighbor r j with the constraint that the two ?? segments?? separated by at least 10 segments along the chain. This are function (Fig. 8) shows a pronounced peak which can be interpreted as the average superhelix diameter. Its position at (18.01.5) nm for 10 mM salt moves to (14.11.5) nm at 20 mM and to (9.41.5) nm at 100 mM salt concentration. There was no further significant change above 100 mM salt concentration. The measured and simulated superhelix diameters were compared above in Fig. 6. The agreement above 10 mM salt concentration is almost perfect. The same results were obtained for p1868. The superhelix diameters simulated for a 7 kb plasmid (Rybenkov et al., 1997a) are shown for comparison; they agree equally well with our experimental data, but they also show a small deviation at 10 mM. In the study by (Rybenkov et al., 1997a), the superhelix diameters were determined via a calculation of the superhelix pitch angle, the writhe and the number of superhelix ends from

11 the simulated conformations. The deviation at low salt concentration in our study might be due to a change in the DNA persistence length, which has not been taken into account in the simulations. To check the influence of the persistence length, we performed a simulation of superhelical pUC18 DNA at 10 mM salt with a persistence length of 75 nm instead of 50 nm, which resulted in a decrease of the superhelix diameter from 18 to 12 nm. Thus, the change in the persiste nce length at this low salt concentration is probably not that large, but it might be responsible for the small deviation of the simulated from the measured superhelix diameter. At 100 mM, the value agrees with the diameter of superhelical regions (9.23.3) nm of p1868 as determined by scanning force microscopy in aqueous solution in the presence of 10 mM MgCl2 and 30 mM NaCl (Rippe et al., 1997).

Fig. 8: Distance distribution function pn(r) of the nearest neighbor of each DNA segment, calculated from and averaged over the simulated configurations of pUC18, at 10 mM ( ), 20 mM ( ), 50 mM ( ) and 100 mM ( ) salt concentration.

4.

CONCLUSIONS

We have studied the salt concentration dependence of the superhelix diameter of two superhelical DNAs p1868 (1868 bp) and pUC18 (2868 bp). The static form factors of both plasmids were measured by small angle neutron scattering in aqueous solution. A sign ificant decrease of the scattering intensity around the scattering vector q=0.5 nm-1 with increasing salt concentration from 0 to 100 mM Na+ in 10 mM Tris was detected. The

12 effect occurred for both plasmids only in the supercoiled form and was independent of DNA concentration. At Na+ concentrations > 0.1 M, no significant change in the scattering curve could be found. Thus, similar to the recent findings of (Gebe et al., 1996), a lateral collapse of the interwound structure for Na + concentrations > 0.1 M, as observed by (Bednar et al., 1994) in cryo-electron microscopy studies, is not supported by our data for superhelical DNA in free solution. The experimental data can be quantitatively reproduced by Monte Carlo simulations. The superhelix diameter, which was here defined as interstrand separation in plectonemic regions of the chain, was calculated from the simulated conformations. It decreases with increasing salt concentration from (18.01.5) nm at 10 mM to (9.41.5) nm at 100 mM. In a recent study, similar values were obtained from MC simulations by a completely different calculation (Rybenkov et al., 1997a). The value at high salt concentration is in agreement with previously reported diameters measured by scanning force microscopy in aqueous solution (Rippe et al., 1997) and with earlier electron microscopy studies (Boles et al., 1990). The used Monte Carlo simulation algorithm reproduces quantitatively the solution struc ture of DNA in solution, it is part of a combined Monte Carlo and Brownian Dynamics algorithm. The Brownian dynamics part has already been shown to calculate correctly certain structural and dynamical features of superhelical DNA, e.g. the salt dependence of internal motions (Hammermann et al., 1997). References
1. Adrian, M., B. ten Heggeler-Bordier, W. Wahli, A. Z. Stasiak, A. Stasiak, and J. Dubochet. 1990. Direct visualization of supercoiled DNA molecules in solution. The EMBO Journal. 9:4551-4554. Bednar, J., P. Furrer, A. Stasiak, J. Dubochet, E. H. Egelman, and A. D. Bates. 1994. The twist, writhe and overall shape of superhelical DNA change during counterion-induced transition from a loosely to a tightly interwound sup erhelix. Possible implications for DNA structure in vivo. Journal of Molecular Biology. 235:825-847. Boles, T. C., J. H. White, and N. R. Cozzarelli. 1990. Structure of plectonemically supercoiled DNA. J. Mol. Biol. 213:931-951. Cantor, C. R., and P. R. Schimmel. 1980. Biophysical Chemistry. Part II: Techniques for the Study of Biological Structure and Function. Freeman, New York. Chirico, G., and J. Langowski. 1996. Brownian dynamics simulations of supercoiled DNA with bent sequences. Biophys J. 71:955-971. Gebe, J. A., J. J. Delrow, P. J. Heath, B. S. Fujimoto, D. W. Stewart, and J. M. Schurr. 1996. Effects of Na+ and Mg2+ on the structures of supercoiled DNAs: comparison of simulations with experiments. J Mol Biol. 262:105-128.

2.

3. 4.

5. 6.

13
7. Hammermann, M., N. Brun, K. V. Klenin, R. May, K. Tth, and J. Langowski. 1998. Salt-dependent DNA superhelix diameter studied by small angle neutron scattering measurements and Monte Carlo simulations. Biophysical J. 75:in press. Hammermann, M., C. Steinmaier, H. Merlitz, U. Kapp, W. Waldeck, G. Chirico, and J. Langowski. 1997. Salt effects on the structure and internal dynamics of superhelical DNAs studied by light scattering and Brownian dynamics. Biophys J. 73:2674-2687. Klenin, K., H. Merlitz, and J. Langowski. 1998. A Brownian dynamics program for the simulation of linear and circular DNA and other wormlike chain polyelectrolytes. Biophys J. 74:780-788. Klenin, K. V., M. D. Frank-Kamenetskii, and J. Langowski. 1995. Modulation of intramolecular interactions in superhelical DNA b curved sequences: a y Monte Carlo simulation study. Biophys J. 68:81-88. Rippe, K., N. Mucke, and J. Langowski. 1997. Superhelix dimensions of a 1868 base pair plasmid determined by scanning force microscopy in air and in aqueous solution. Nucleic Acids Res. 25:1736-1744. Rybenkov, V. V., A. V. Vologodskii, and N. R. Cozzarelli. 1997a. The effect of ionic conditions on the conformations of supercoiled DNA .1. Sedimentation analysis. J Mol Biol. 267:299-311. Rybenkov, V. V., A. V. Vologodskii, and N. R. Cozzarelli. 1997b. The effect of ionic conditions on the conformations of supercoiled DNA . 2. Equilibrium catenation. J Mol Biol. 267:312 -323.

8.

9.

10.

11.

12.

13.